Transcriptome study of human embryonic stem cells and knockdown study of a pluripotency marker, LIN28

Functional analysis of LIN28’s role in stem cell pluripotency was conducted by siRNA- and shRNA-mediated LIN28 knockdown followed by gene expression profiling using Illumina microarrays

A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF BIOLOGICAL SCIENCES NATIONAL UNIVERSITY OF SINGAPORE

2010

ACKNOWLEDGEMENTS

I would like to express my deepest appreciation to my thesis supervisor, Associate Professor Chan Woon Khiong, for his persistent patience, support, and dedication in guiding me to accomplish this research project

I would also like to thank Dr Shubha Vij for her invaluable guidance and advice in SAGE data analysis Thanks are given to Wang Yue for her assistance in human embryonic stem cell culture and providing embryoid bodies Special thanks go to Chak Li Ling, Tan Jee Hian, Allan and to Dr Maria D Serafica for their invaluable assistance in Illumina Microarray experiment and data analysis

I am also grateful to people from Molecular Genetics Laboratory, especially Dr Geeta Ravindran, Dr Shenoy Sudheer, Pham Nguyet Minh, and Wong Pui Mun I would like to thank all my other friends here at NUS, for their friendships and companionship, which have given me confidence to accomplish my project

Last but not least, my deepest thanks go to my parents for their great love, incredible trust, and constant support, which is the driving force behind the successful completion of

my research

1.1 Human embryonic stem cells

1.1.1 Overview and characteristics of human embryonic stem cells

1.1.2 Regulatory networks and transcription factors in human ES cells

1.1.3 Induced pluripotent stem cells

1.1.4 Human embryonal carcinoma cells

1.2 Transcriptome studies of human ES cells

1.2.1 DNA Microarray

1.2.2 Expressed Sequence Tags Scan

1.2.3 Massively Parallel Signature Sequencing

1.2.4 Serial Analysis of Gene Expression

1.3 RNA interference in human ES cells

1.4 LIN28 is an important regulator for pluripotency in human ES cells

1.4.1 The interaction of LIN28 and microRNA let-7 family is important

for human ES cells

1.4.2 LIN28 can regulate target genes post-transcriptionally

1.4.3 Overexpression and knockdown studies of LIN28

1.5 Objective of the current study

I II V VII VIII IX

2 Materials and methods

2.1 Culture of human ES cell line

2.1.1 Preparation of feeder cells

2.1.2 Maintenance of human ES cells

2.1.3 Preparation of embryoid bodies

2.2 Culture of NCCIT cell line

2.3 Preparation of shRNA vectors targeting LIN28

2.4 Transfection and lentivirus transduction of mammalian cells

2.4.1 Transfection of supercoiled shRNA vectors

2.4.2 Transfection of siRNA

2.4.3 Lentivirus transduction of NCCIT

2.4.4 Lentivirus transduction of HES3

2.5 Fluorescence Activated Cell Sorting (FACS)

2.6 SAGE data analysis

2.6.1 SAGE Libraries

2.6.2 Pair-wise comparison

2.6.3 Hierarchical Clustering Analysis and Transchisq clustering

2.7 Illumina Microarray

2.7.1 Isolation of total RNA

2.7.2 Synthesis of double-stranded cDNA and amplification of cRNA

2.7.3 Hybridization, wash and scan of Illumina microarray

2.7.4 Bioinformatics data analysis of Illumina microarray

2.8 Quantitative Real-time PCR (qRT-PCR)

3 Results

3.1 SAGE data analysis to search for pluripotency and differentiation markers

3.1.1 Pair-wise comparisons among human ES cell SAGE libraries

3.1.2 Hierarchical Cluster Analysis of SAGE data

3.1.3 Transchisq analysis identified gene expression patterns during

human ES cell differentiation

3.1.4 The clustered gene differential expression patterns were

3.2 LIN28 knockdown study

3.2.1 Lin28 transient knockdown in NCCIT

3.2.2 Lin28 shRNA conditional stable line construction

3.2.3 Microarray data analysis

3.2.3.1 Effect of cationic lipid-based transfection reagents on

NCCIT expression profile 3.2.3.2 Effect of EGFP expression on NCCIT expression profile

3.2.3.3 Effect of transient LIN28 knockdown on gene

expression profile of NCCIT

4 Discussion

4.1 SAGE data analysis provides robust candidates for stemness assessment

4.1.1 Hierarchical Cluster Analysis identifies a major ES/EC-specific

cluster

4.1.2 Genes co-expressed with POU5F1, SOX2 and NANOG possess

binding sites for these core pluipotency factors

4.1.3 Transchisq clustering reveals new potential pluripotency and

differentiation markers based on expression pattern

4.2 LIN28 knockdown reveals its role in pluripotency at the

post-transcriptional level

4.2.1 Cationic lipid-based transfection reagents deliver DNA into cells

through endocytosis and bring toxicity

4.2.2 EGFP should not be considered as a biologically inert indicator

4.2.3 LIN28 knockdown does not cause differentiation of pluripotent

stem cells

4.2.4 LIN28 is an essential factor in post-transcriptional regulation

4.2.5 LIN28 regulates other RNA post-transcriptional regulators

4.2.6 Inducible NCCIT LIN28sh stable line can be a good tool for

embryonic development study

4.3 Conclusions and future work

SUMMARY

The amount and the pace of research on human embryonic stem (ES) cells is currently going on at an unprecedented rate due to their potential as a limitless source of cells for regenerative medicine and cellular repair The key to utilizing the regenerative capability of human ES cells lies in elucidating the mechanisms underlying self-renewal and pluripotency, the two defining features of human ES cells

We compared in-house human ES cell SAGE libraries with other ES cell, embryonal carcinoma (EC) cell, cancer and normal tissue SAGE libraries available in public databases

A major ES/EC cluster was identified using Hierarchical Clustering Analysis Potential pluripotency gene markers were identified as such because they shared the same gene expression profile with well-known pluripotency markers like POU5F1/LIN28, SOX2 and NANOG A Transchisq algorithm-based clustering method identified gene expression patterns upon differentiation of HES3 cells These patterns were validated by quantitative real-time PCR (qRT-PCR) analysis For the qRT-PCR confirmation, instead of taking two extreme data sets such as undifferentiated and a late stage embryoid body, a time series of embryoid body stages ranging from 12h to 14 days was profiled Based on both the SAGE data and experimental qRT-PCR data, we proposed TERF1, SOX2, C14ORF115, NANOG and LIN28 could be the good pluripotency markers and differentiation marker such as DCN, AA853630 and APOC3 could serve better to assess the true state of the pluripotent cells, due to their earlier and higher fold change in expression upon differentiation

LIN28, one of the four factors sufficient to reprogram adult fibroblast cells into induced pluripotent stem (iPS) cells, plays important roles in embryonic development Functional analysis of LIN28’s role in stem cell pluripotency was conducted by siRNA- and shRNA-mediated LIN28 knockdown followed by gene expression profiling using Illumina microarrays in human embryonal carcinoma (EC) cell line, NCCIT, which was

used as alternative model to human ES cells because of its resemblance to human ES cells and its convenience for culture After knockdown, none of the genes involved in pluripotency or differentiation showed significant change of expression A set of genes related to various post-transcriptional regulatory steps such as mRNA splicing, cytoplasmic polyadenylation, and mRNA stabilization were identified We proposed that LIN28 might act as a master regulator in differentiation and establishment of pluripotency

by directly modulating genes responsible for pluripotency or by modulating other transcriptional regulators to form a hierarchical post-transcriptional control A conditional LIN28 knockdown stable line was established from NCCIT, which could be a good tool to study the LIN28’s role in pluripotency and differentiation

post-LIST OF TABLES

Table 1 Oliogonucleotides used in shRNA vector cloning 24 Table 2 SAGE libraries used in this study 29 Table 3 List of primers used in qRT-PCR (SYBR Green Assay) 35 Table 4 List of genes bound by pluripotency transcription factors 41

Table 5 List of genes selected for confirmation of expression

Table 6

List of top 10 enriched biological process GO terms from common genes affected by both FuGENE HD and RNAiMAX

56

Table 7 List of top 10 enriched biological processes GO terms from

genes affected by EGFP 57 Table 8 List of top 10 enriched biological process GO terms from

genes differentially expressed after LIN28sh knockdown 60 Table 9 List of genes differentially expressed after LIN28sh

knockdown related to RNA binding or RNA processing 61

LIST OF FIGURES

Figure 1 ES cells’ two defining features: self-renewal and

Figure 2 Regulatory networks and transcription factors in maintenance

Figure 3 Lentirivirus-mediated shRNA knockdown 15

Figure 4 Construction of lentiviral inducible shRNA vectors targeting

Figure 5 Comparisons of different SAGE libraries using

DiscoverySpace software 37

Figure 6 Hierarchical Cluster Analysis of human ES/EC libraries with

normal and cancer tissue/cell lines 40 Figure 7 Transchisq clustering of undifferentiated, partially differentiated and differentiated HES3 cells 43

Figure 8 qRT-PCR to confirm the expression patterns during differentiation clustered by Transchisq clustering 47 Figure 9 Transient transfection of pLVET LIN28sh vectors into NCCIT cell line 49 Figure 10 siRNA knockdown of LIN28 in NCCIT cells 50 Figure 11 pLVET LIN28sh lentivirus transduction on HES3 and NCCIT cells 51

Figure 12 Inducibility test on NCCIT LIN28sh stable line 53

Figure 13 Expression profile affected by cationic lipid-based

LIST OF ABBREVIATIONS

Abbreviation Meaning

APOC3 Apolipoprotein C-III

Blimp B lymphocyte induced maturation protein

BMP bone morphogenetic protein

C14ORF115 Chromosome 14 open reading frame 115

ChIP chromatin immunoprecipitation

DCN Decorin

DNMT3B DNA (cytosine-5-)-mythyltransferase 3 beta

DOX doxycycline

dsRNA double-strand RNA

EGFP enhanced green fluorescent protein

ERK Extracellular signal-regulated kinase

ES cells Embryonic stem cells

EST Expressed sequence tags

FACS Fluorescence Activated Cell Sorting

HMGA2 High-mobility group AT-hook 2

GCTs Germ cell tumors

GIS Gene Identification Signature

GLGI Generation of Longer cDNA fragments from SAGE tags for Gene

Identification HCA Hierarchical Cluster Analysis

hdFs human-ES-cell-derived fibroblast-like cells

IGF-2 insulin-like growth factor-2

iPS cells induced pluripotent stem cells

Klf4 Kruppel-like factor 4 (gut)

KRAB Kruppel-associated Box gene

LIF leukemia inhibitory factor

Abbreviation Meaning

miRNA microRNA

MOI multiplicity of infection

MPSS Massively Parallel Signature Sequencing

MYC v-myc myelocytomatosis viral Oncogene homolog NANOG Nanog homeobox

NATs natural antisense transcripts

NODAL nodal homolog (mouse)

NTC non target control

PETs paired-end ditags

PGCs primordial germ cells

PI3K phosphoinositide-3-kinase POU5F1 POU class 5 homeobox 1

qRT-PCR quantitative real-time PCR

RBPs RNA binding proteins

REX1 RNA exonuclease 1 homolog

RHA RNA helicase A

RISC RNA-induced silencing complexes

RNAi RNA interference

RNPs ribonucleoprotein particles SAGE Serial Analysis of Gene Expression

shRNAs short hairpin RNAs

siRNA small interfering RNA

SNP single nucleotide polymorphisms

SOX2 SRY (sex determining region Y)-box 2

SSEA-3 stage-specific embryonic antigen-3

TERF1 Telomeric repeat-binding factor 1

tetR tetracycline repressor

TGF transforming growth factor

TPM tag per million

TRA tumor rejection antigens

UTR untranslated region

WNT wingless-type MMTV integration site family

Chapter 1 Introduction

1.1 Human embryonic stem cells

1.1.1 Overview and characteristics of human embryonic stem cells

Embryonic stem (ES) cells are isolated from the inner cell mass (ICM) of embryos of

blastocyst stage (Martin, 1981; Evans and Kaufman, 1981; Thomson et al., 1998; Reubinoff et al., 2000) Research on ES cells could be traced back to 1950s with the study

of germ cell tumors identified as teratocarcinomas Later, in the 1970s, the embryonal

carcinoma (EC) cell line was isolated from teratocarcinomas and cultured in vitro permanently (Jakob et al., 1973; Gearhart and Mintz, 1974) The pioneering work in

mouse EC cells paved the way to the derivation of pluripotent cells from the ICM of mouse blastocysts, termed embryonic stem (ES) cells, under culture condition of fibroblast feeder layers and serum (Martin, 1981) Since then, efforts have been undertaken to establish

human ES cells Bongso et al (1994) first reported the primary cultures of undifferentiated

cells from the human blastocyst These cells eventually underwent differentiation or death,

as they relied on leukemia inhibitory factor (LIF) supplementation of the culture medium instead of embryonic feeder cell support In 1998, Thomson and co-workers (1998) reported the successful establishment of human ES cell line from blastocysts

Pluripotency and self-renewal are the two defining features of ES cells (Fig 1) renewal is defined by the ES cells’ capability to proliferate permanently without differentiating under culture conditions Pluripotency refers to the potential which ES cells possess to differentiate into all kinds of cell types, basically including three germ layers, endoderm (interior stomach lining, gastrointestinal tract, the lungs), mesoderm (muscle, bone, blood, urogenital), or ectoderm (epidermal tissues and nervous system) Traditionally,

Self-ICM cells were thought to be excluded from the trophectoderm lineage (Beddington and Robertson, 1989) However, it was subsequently found that the ICM still possess the

ability to differentiate into the trophectoderm lineage (Pierce et al., 1988), and mouse ES cells can also be differentiated in trophectoderm under certain culture condition (Niwa et al., 2005) Some researchers have redefined pluripotency as the ability to generate all cell

types including trophectoderm but without the self-organizing ability to develop into a whole embryo (Solter, 2006; Niwa, 2007) Although these two characteristics describe different aspects of ES cell, they are closely related to each other For instance, ES cell pluripotency is maintained via self-renewal by the prevention of differentiation and the promotion of proliferation under proper culture conditions (Niwa, 2007)

Figure 1 ES cells’ two defining features: self-renewal and pluripotency Under certain condition ES cells

can proliferate permanently Meanwhile, ES cells possess the potential to differentiate into all cell types (endoderm, mesoderm, ectoderm and trophectoderm) but without the self-organizing ability to develop into a whole body

Because of their unlimited proliferation and capability to contribute to any tissue, human ES cells are considered as an unprecedented source of cells for potential therapy for

a wide range of degenerative diseases (Wobus and Boheler, 2005; Hyslop et al., 2005a)

After directed differentiation into target functional somatic cells, purification and transplantation, ES cells have already been proven to contribute to the recovery from post-

infarction syndrome (Min et al., 2002), Parkinson’s disease (Kim et al., 2002), Huntington’s disease (Dinsmore et al., 1996), and diabetes (León-Quinto et al., 2004) in

animal models

1.1.2 Regulatory networks and transcription factors in human ES cells

Various signaling pathways appear to be responsible for maintenance of human ES cells (Fig 2) Unlike mouse ES cells, the combination of LIF and bone morphogenetic protein (BMP)-4 is not sufficient to maintain human ES cells On the contrary, BMP-4

causes their differentiation towards trophectoderm (Xu et al., 2002; Gerami-Naini et al., 2004; Bai et al., 2010) In contrast to BMP-4, other transforming growth factor (TGF) - β

family members such as Activin A, TGFβ1 and Nodal appear to promote pluripotency in human ES cells, through the activation of Smad 2/3 that subsequently induces the expression of Nanog homeobox (NANOG) and POU class 5 homeobox 1 (POU5F1)

(Vallier et al., 2005; Babaie et al., 2007) For human ES cells, basic fibroblast growth factor (bFGF) is an indispensable component (Amit et al., 2000) Recently, Bendall et al

(2007) elucidated that pluripotency of human ES cells is dependent on their interplay with human-ES-cell-derived fibroblast-like cells (hdFs), involving bFGF and insulin-like growth factor-2 (IGF-2) signaling Activated by IGF pathway, Phosphoinositide-3-kinase

(PI3K) (Sato et al., 2004; McLean et al., 2007; Hui et al., 2010) and Extracellular regulated kinase (ERK) (Li et al., 2004; Feng, 2007; Wang et al., 2010) signalings have

signal-been proven to be crucial for human ES cells self-renewal Another important pathway is canonical wingless-type MMTV integration site family (WNT) signaling, which is sufficient to maintain self-renewal of human ES cells and through its downstream

components β-Catenin, it can sustain the expression of POU5F1 and NANOG (Sato et al., 2004; Ogawa et al., 2006)

Transcription factors play essential roles in the maintenance of pluripotency in human ES cells The best studied is POU5F1, also known as OCT4 or OCT3, which encodes a POU domain factor The balance of POU5F1 expression level is very important

to the maintenance of pluripotency When POU5F1 is overexpressed, human ES cells will develop into endoderm; nevertheless, when it is lost, human ES cells will be directed into

trophectoderm and primitive endoderm (Hay et al., 2004; Rodriguez et al., 2007; Babaie et al., 2007) SRY (sex determining region Y)-box 2 (SOX2), another important transcription

factor, is known to cooperate with POU5F1 to form POU5F1-SOX2 complex to activate

the target genes in a synergistic way (Chew et al., 2005) Knockdown of SOX2 in human

ES cells resulted in loss of the undifferentiated stem cell state, as indicated by a change in cell morphology, reduced expression of key stem cell factors and increased expression of

trophectoderm markers (Fong et al., 2008) Knockdown of NANOG by small interfering

RNA (siRNA) can lead human ES cells differentiation towards extraembryonic lineages

(Hyslop et al., 2005b) Zaehres et al., (2005) using a NANOG RNA interference (RNAi)

stable line, reported that NANOG had an antagonizing role in endodermal and

trophectodermal differentiation Boyer et al (2005) using chromatin immunoprecipitation

(ChIP) combined with genome-wide location techniques, showed that POU5F1, SOX2, and NANOG shared a large number of target genes in active or inactive status Based on these results, they proposed that these transcription factors form a regulatory circuitry consisting of autoregulatory and feedforward loops to maintain the pluripotency in human

ES cells

Figure 2 Regulatory networks and transcription factors in maintenance of human ES cells bFGF is an

essential component in human ES cell culture, which binds to human-ES-cell-derived fibroblast-like cells (hdFs) to promote its IGF2 secretion IGF2 signaling promotes pluripotency through PI3K/ERK pathway Unlike mouse ES cells, BMP appears to inhibit pluripotency by phosphorylating Smad 1/5/8 in human counterparts WNT, TGF β and Activin A are proven to promote OCT4 and NANOG expression Three core transcription factors, OCT4, SOX2, and NANOG share a large number of target genes and form a regulatory feedback circuit to maintain pluripotency

1.1.3 Induced pluripotent stem cells

In 2006, a Japanese group succeeded in generating mouse induced pluripotent stem (iPS) cells from mouse fibroblasts (Takahashi and Yamanaka, 2006) using only four transcription factors: Pou5f1, Sox2, c - v-myc myelocytomatosis viral Oncogene homolog (c-Myc), Kruppel-like factor 4 (gut) (Klf4) These iPS cells are highly similar to ES cells

in terms of self-renewal and pluripotency, and they are proven to be able to generate all

cell types (Maherali et al., 2007; Okita et al., 2007) Later, they achieved generation of human iPS cells using the same four factors (Takahashi et al., 2007) Meanwhile, another

group from the U.S also reported the successful generation of human iPS cells where they also used POU5 and SOX2 shared by the previous reprogramming gene panel, but replaced

MYC and KLF4 with NANOG and LIN28(Yu et al., 2007)

Extensive efforts have been taken to improve the reprogramming system One direction is to minimize the gene set to reprogram Oncogene c-Myc is proven to be dispensable for reprogramming for both mouse and human fibroblasts with lower

efficiency (Nakagawa et al., 2008) The orphan nuclear receptor Esrrb, incorporated with Oct4 and Sox2 can accomplish mouse reprogramming (Feng et al., 2009) It has been

reported that two factors (Oct4 and Klf4 or c-Myc) are sufficient to reprogram mouse

neuronal progenitors (Kim et al., 2008) Even Oct4 alone can generate iPS cells from adult mouse neural stem cells in spite of low efficiency (Kim et al., 2009a) Another direction of

improvement is to reduce genome integration events related to tumorigenesis integrating adenoviral system was employed successfully to generate human iPS cells

Non-(Zhou and Freed, 2009) Other non-integrating viruses such as Sendai virus (Fusaki et al., 2009) and Epstein-Barr virus (Yu et al., 2009) are also able to generate human iPS cells

and the transgenes were lost gradually after reprogramming A single viral vector carrying all four reprogramming factors was used to generate mouse and human iPS cells through

only one genome integration (Carey et al., 2009) Using piggyback transposon, Kaji et al

(2009) induced virus-free iPS cells with subsequent excision of the reprogramming factors Without any virus integration and modification of the target genome, two studies provided safer manners to generate iPS cells Consecutive transfections of RNA were carried out to support continuous protein expression of four core reprogramming factors, which resulted

in iPS cell colonies from human fibroblasts successfully (Yakubov et al., 2010) Delivery

of recombinant reprogramming proteins has been reported to generate mouse iPS cells too

(Zhou et al., 2009) All these researches have explored the therapeutic potential of iPS as

patient-specific and genetically compatible cell sources to a large extent

1.1.4 Human embryonal carcinoma cells

Germ cell tumors (GCTs) arise from primordial germ cells (PGCs) Within GCT category, seminomas are generally histologically uniform and seem to resemble a transformed state of the PGC Nonseminomatous GCTs, on the other hand, typically include teratocarcinomas with EC components, which are considered as the ‘pluripotent’

stem cells of these cancers (Sperger et al., 2003) Despite their germ cell origin, EC cells

share many commonalities with ES cells in various aspects Like ES cells, EC cells

proliferate extensively both in vitro and in vivo and have the potential to differentiate into cell types from all three germ layers (Andrews et al., 1984a) If injected into the inner cell

mass of early embryos, EC cells can contribute to generating chimeric mice as well (Mintz and Illmensee, 1975) Both cells express the core stemness transcription factors, POU5F1,

SOX2, and NANOG, controlling the undifferentiated state (Sperger et al., 2003; Boyer et al., 2005) Compared to human ES cells, the most significant difference of human EC cells

is their karyotypical aberration (Wang et al., 1980), such as acquirement of additional copies of chromosome 17 and chromosome 12 (Rodriguez et al., 1993; Skotheim et al.,

2002)

The tumorogenic potential of human EC cells makes them unusable for future regenerative medicine, but they are a good model to study pluripotency and early

embryonic development (Josephson et al., 2007) Human EC cells have the following

major advantages: compared to human ES cells, they can grow without the support of feeder layers; they are easy to passage; they are resistant to spontaneous differentiation; they are widely available without intellectual property restraints and burdensome

regulations (Josephson et al., 2007) Many pluripotency markers were originally

discovered as antigens of human EC cells These markers include stage-specific embryonic

antigen-3 (SSEA-3) (Shevinsky et al., 1982; Damjanov et al., 1982), SSEA-4 (Kannagi et

al., 1983), and tumor rejection antigens (TRA)-1-60 and TRA-1-81 (Andrews et al.,

1984b)

Based on transcriptome studies, it has been shown that the ES cells and EC cells

share similar overall gene expression profiles (Sperger et al., 2003; Liu et al., 2006) A

microarray study using various human ES cell lines and human GCTs highlighted a set of

565 genes highly expressed in ES cells and EC cells but not in seminomas (Sperger et al.,

2003) This supports the hypothesis that seminomas closely resemble transformed PGCs, while EC cells mostly represent a reversion to more ICM- or primitive ectoderm-like cells

Similarly, Liu et al (2006) also showed that EC cells are clustered together with ES cells

while differentiated EC cells and embryoid bodies (EBs) can be readily distinguished from their parent populations

1.2 Transcriptome studies of human ES cells

1.2.1 DNA Microarray

DNA microarray is a multiplex detection and characterization technology based on DNA and complementary DNA (cDNA) or complementary RNA (cRNA) hybridization A large number of cDNA or oligonucleotides are spotted on membranes, glass surface or plastic as unique probes to achieve a high throughput screening DNA microarray has

become one of the main platforms for genome wide expression analysis (Schena et al., 1995; Noordewier and Warren 2001; Holloway et al., 2002)

DNA microarray has been widely used in exploring human ES cells stemness signature In one of these early studies, 918 genes enriched in undifferentiated human ES cell line H1 compared with their non-lineage directed differentiated counterparts were

identified (Sato et al., 2003) Recently, the Illumina BeadArray microarray platform has

also been found popular as its advantages include high sensitivity, redundance of technical

replicates, smaller sample sizes and the ability of running samples simultaneously Liu et

al., (2006) profiled 48 different samples, including human ES cells, EBs differentiated

from them, karyotypically abnormal human ES cell line BG01V, human fibroblast feeder and EC lines using Illumina BeadArray Another group used BeadArray to study

transcriptome co-expression map of human ES cells (Li et al., 2006) Among the total 754

co-expression domains identified from ES and EB expression data, only 18 domains were shared by ES and EB, indicating that the co-expression maps were different between them This study initiated the examination of how transcriptional regulation interacts with genomic structure and how genes clustered on the same chromosome are co-expressed during the ES cells self-renewal and differentiation

1.2.2 Expressed Sequence Tags Scan

Expressed sequence tags (EST) scan is a technology based on single-pass sequencing

of cDNAs (Parkinson and Blaxter, 2009; Clifton and Mitreva, 2009) In the beginning of human genome project, EST scan was the main method to profile various tissues and discover novel transcripts Two extensive EST analyses of human ES cells were reported

by Brandenberger et al (2004a) and Miura et al (2004) In the former study, 148,453 high

quality ESTs (32,764 unique transcripts) were obtained, in which 52% of unique transcripts could not be mapped to a UniGene transcripts and represented potentially novel genes Human ES cell EST data was also compared with that of three partially differentiated cell populations derived from different protocols, thus increasing reliability

of differentially expressed gene list A total of 672 differentially expressed genes were identified, and of these, 70% were validated to be differentially expressed by qRT-PCR

(Brandenberger et al., 2004a) This study also highlighted differentially genes in respect of

important signaling pathways related to stem cell maintenance While LIF signaling components were not detected, all FGF receptors were up-regulated in undifferentiated ES

cells There were also many WNT and nodal homolog (mouse) (NODAL) pathway components, both agonists and antagonists in the list, suggesting that they were tightly controlled for proper growth and differentiation of human ES cells In another study, three different ES cell lines (H1, H7 and H9) and their 14-D EBs were used to generate EST data,

to monitor the state of human ES cells derived from different laboratories using

independent methods and maintained under various culture conditions (Miura et al., 2004)

In this study, in addition to discovery of novel plupotency genes, pathways such as WNT and TGFβ were stressed in the maintenance of pluripotency

1.2.3 Massively Parallel Signature Sequencing

Massively Parallel Signature Sequencing (MPSS) is comprised of two steps: a) in vitro cloning of cDNA fragments tagged by DpnII on microbeads and b) several rounds of

ligation-based sequencing Typically, a sequence signature of 17 bp is determined

representing its corresponding mRNA molecules (Brenner et al., 2000) In each experiment,

over a million signature sequences can be generated in parallel, reaching sensitivity at a level of a few molecules of mRNA per cell

Wei et al., (2005) utilised MPSS to study human ES cell transcriptome In this study, two human ES cell lines were compared with one mouse ES cell line The results showed that only a small core set of genes were shared by both types of ES cells compared to differentiating EBs, while a large number of differences was observed indicating the cross species biological pathway distinctions They also pointed out that tags containing a double palindrome or falling in a repeat region (eg Human NANOG and RNA exonuclease 1 homolog (REX1)) could not be detected Brandenberger and colleagues (2004b) used MPSS to identify eleven thousand unique transcripts from pooled H1, H7, and H9 undifferentiated human ES cells, of which approximately 25% were novel transcripts The

top 200 abundant transcripts constituted 99% of the total number of counts, among which there were only three known ES cell markers, namely SOX2, DNA (cytosine-5-)-mythyltransferase 3 beta (DNMT3B) and OCT4 Most of the top 200 genes were ribosomal genes or genes related to protein and nucleic acid synthesis No expression bias

of chromosomal regions was observed and genes from both X and Y chromosomes were

detected Similar to the findings from EST study (Brandenberger et al., 2004a),

components of signaling pathways were detected but their inhibitors were also present, indicating the role of negative regulation in maintaining the pluripotency state

(Brandenberger et al., 2004b)

1.2.4 Serial Analysis of Gene Expression

Serial Analysis of Gene Expression (SAGE) is another popular method in transcriptome study Conventional SAGE protocol produces 14-nucleotides tags to represent an individual transcript Like MPSS, SAGE allows quantitative characterization

of the transcriptome and has advantages over microarray in its ability to identify novel

splice variants, exons and genes (Velculescu et al., 1995)

SAGE cannot reach the depth of MPSS data and its standard cloning and sequencing are labor-consuming, but MPSS’s high cost and requirement of complex facility prevent researchers from smaller labs from choosing it It has been reported that the SAGE is 26 times more sensitive than the EST method for the detection of low abundance transcripts

(Sun et al., 2004) However, in spite of its great sensitivity, SAGE method suffers from

ambiguity because of its short sequence signature In one report, about half of the SAGE tags could not match any known expressed sequences and more than one third of the

SAGE tags that mapped to known expressed sequences, had multiple matches (Chen et al.,

2002) Additionally, during the annotation, the short tags require 100% match to the public

available reference databases (SAGEmap (Lash et al., 2000) or SAGE Genie (Boon et al.,

2002), making the method more susceptible to single nucleotide polymorphisms (SNP), PCR and sequencing errors All these drawbacks are adversely influencing the power of SAGE technology

Many efforts have been taken to reduce the ambiguity of SAGE Using MmeI (LongSAGE) or Ecop15I (SuperSAGE) as tagging enzymes instead of BsmFI (SAGE), the tag length can be increased to 21 or 27 bp respectively (Saha et al., 2002; Matsumura et al.,

2003) Nevertheless, LongSAGE protocol generates two-nucleotide recessed 5’ ends, which are not filled, thus compromising the faithfulness of transcriptome profiling The

unpredictability of Ecop15I has also inhibited its application in complex genome like human genome Reverse SAGE (rSAGE) (Yu et al., 1999) or Generation of Longer cDNA fragments from SAGE tags for Gene Identification (GLGI) (Chen et al., 1999) have been

developed to explore the novel genes or the ones with ambiguous tag identity Another

strategy called Gene Identification Signature (GIS) has been developed (Ng et al., 2005), whereby MmeI cuts 18bp signature pairs from the 5’ and 3’ ends of full length cDNA for

gene annotation, rather than a single SAGE tag They demonstrated that 95.2% of 34,815 single-locus paired-end ditags (PETs) had matches to known transcripts

The first SAGE analysis of the human ES cells was conducted in HES3 and HES4

lines with different genetic and ethnic backgrounds (Richards et al., 2004) The overall

profiles of HES3 and HES4 showed basic similarity Most abundant genes were involved

in DNA repair, stress responses, apoptosis, cell cycle regulation and development Seventy-three ribosomal proteins were more abundant than in normal tissues The differences between HES3 and HES4 were attributed to different gender backgrounds amongst other factors Comparison of the human ES cells SAGE data with the 21 SAGE libraries from normal and cancer tissues not only confirmed known ES-specific markers like POU5F1, SOX2, NANOG and REX1, but also highlighted some other less well

characterized transcription factors including LIN28 and DNMT3B, which were validated

by subsequent transcriptome studies repeatedly (Brandenberger et al., 2004a; Hirst et al., 2007; Assou et al., 2007) Moreover, LIN28 recently was proven to be one of four potent

ES cells factors sufficient to reprogram the somatic cells into pluripotent stem cells (Yu et al., 2007) The authors also compared their SAGE data with available mouse ES cell

SAGE data and concluded that regardless of basic similarities between human and mouse

ES cells, there were significant differences in their respective regulatory pathways Hirst et al., (2007) reported similar gene expression profiles among nine human ES cell derived

from different sources, using longSAGE protocol In this study, they found increased expression of transcripts for RNA binding proteins in human ES cells compared to four terminally differentiated cells and 52 novel apparently ES-specific tags were extended by 5’ RACE, the majority of which represented non-coding RNAs In order to convert

“orphan” tags into more useful information, Richards et al (2006) chose rSAGE to convert

“orphan” tags into more useful information This study proved that the SNPs had a significant impact on the correct assignment of SAGE tags Furthermore, the rSAGE approach was shown to be useful in identification of natural antisense transcripts (NATs), novel introns and new splice variants of known transcripts

1.3 RNA interference in human ES cells

RNA interference (RNAi) is a post-transcriptional gene regulatory mechanism which represses the transcript level inside the living cells RNAi is evolutionarily conserved in a wide range of eukaryotes including animals (Siomi and Siomi, 2009) The RNAi reaction

is initiated by the enzyme Dicer, which cleaves the double-strand RNA (dsRNA) into

21-25 bps short fragments One of the two strands of each fragment, known as the guide strand, is then incorporated into the RNA-induced silencing complex (RISC) Subsequently,

this complex binds to the target mRNAs matched by the guide strand and cleave the mRNAs or repress their transcription (Hannon, 2002) The two types of central molecules involved in RNAi mechanism are microRNA (miRNA) and small interfering RNA (siRNA) Typically, miRNAs are derived from endogenously expressed precursor RNAs and they interact with target mRNAs by recognizing their 3’ untranslated region (UTR)

(Lagos-Quintana et al., 2002) On the other hand, siRNAs are produced from DNA templates expressing short hairpin RNAs (shRNAs) (Paddison et al., 2002)

The specificity and robust efficiency of RNAi on gene expression make it a valuable research tool in cell lines and in living organisms (Fig 3) Compared with other silencing methods such as antisense oligonucleotides and ribozymes, RNAi tends to be more

effective and less toxic (Miyagishi et al., 2003) In order to achieve RNAi, chemically

synthesized siRNA molecules or a plasmid producing shRNA can be used to transfect cells Usually, RNA polymerase type III promoters such as the U6 small nuclear RNA promoter

or H1 promoter are used to drive shRNA expression from the template vectors (Paul et al., 2002; Brummelkamp et al., 2002) Despite being quick, convenient and cost-effective,

siRNA and shRNA plasmid transfection remain limited because of the transient nature of expression and variable transfection efficiencies To overcome these drawbacks, virus-based high-efficiency shRNA delivery systems have been developed (Devroe and Silver,

2002; Xiong et al., 2005) However, constitutive expression of shRNA cannot be used if a

gene functions during multiple critical development stages Thus, drug-controllable RNAi has also been developed, which allows for conditional knockdown of endogenous genes

(Szulc et al., 2005; Matthess et al., 2005)

Figure 3 Lentirivirus-mediated shRNA knockdown Vectors containing shRNA are packaged into

lentiviral particles, which are then transduced into mammalian cells Next, the fragments carrying shRNA are integrated into the genome of target cells as templates to express shRNAs After Drosha processing, shRNAs are transported into cytoplasm and cleaved into siRNA by Dicer One strand (guide strand) of the double- strand siRNA is associated with RISC to either cleave or repress the transcription of target mRNA matched

by the guide strand [From Dr Dan Cojocari’s web page, Department of Medical Biophysics, University of Toronto 2010]

The use of RNAi to knockdown the expression of genes suspected to be functionally important for maintenance of human ES cells has facilitated efforts aimed at key genes involved in self-renewal and pluripotency, including OCT4, NANOG, SOX2, LIN28, Zic

family member 3 (ZIC3) (Hay et al., 2004; Hyslop et al., 2005b; Lim et al., 2007; Rodriguez et al., 2007; Fong et al., 2008) In mouse ES cells, a cDNA-based RNAi library has been generated to facilitate high-throughput functional genetic screens (Jian et al.,

2007) The system used a vector with a convergent H1 and U6 dual promoter for the

expression of dsRNA from randomly inserted cDNA Because RNAi-mediated knockdown

of specific genes in human ES cells promotes their differentiation towards specific lineages,

it can be a potential genetic tool to obtain desired cell types from human ES cells, which can be good sources for cell therapy for degenerative diseases (Rassouli and Matin, 2009)

1.4 LIN28 is an important regulator for pluripotency in human ES cells

LIN28, an RNA binding protein, contains three RNA binding domains: one shock domain at the N terminus and two CCHC-type zinc finger domains at the C terminus

cold-It was originally found to regulate developmental timing in C elegans (Ambros and

Horvitz, 1984)

Consistent with its function in C elegans, mammalian LIN28 is found to be

expressed in embryonic muscle, neurons, and epithelia in a stage-specific manner

(Polesskaya et al., 2007) In addition, LIN28 is specifically expressed in undifferentiated

ES cells and EC cells and is reduced during differentiation (Richards et al., 2004; Polesskaya et al., 2007) Recent achievement of iPS cells, by overexpressing four genes

containing LIN28, reinforced the notion that LIN28 is a key regulator of pluripotency in

human ES cells (Yu et al., 2007)

1.4.1 The interaction of LIN28 and microRNA let-7 family is important for pluripotency in human ES cells

Earlier, mouse Lin28 was reported to be regulated by microRNA miR-125b (a mammalian microRNA homologous to Lin4) post-transcriptionally accompanied by cell

differentiation (Wu and Belasco, 2005) In C elegans, Lin4 binds imperfectly to

complementary sites in the 3'-UTR and inhibits translation of target mRNAs, including

Lin28, Hbl-1, and Lin14 (Lin et al., 2003), among which Lin14 protein is able to repress

the Lin28 translation proceeding after initiation (Seggerson et al., 2002) Thus, Lin28 may

be regulated by miR-125b directly and indirectly

Several recent papers reported that LIN28 represses let-7 microRNA family maturation, but the mechanism remains largely unknown Several studies found that mouse Lin28 binds to the terminal loop regions of let-7 precursor pri-let-7 and represses the let-7

microRNA at the Drosha processing stage (Newman et al., 2008; Piskounova et al., 2008; Viswanathan et al., 2008) However, another group demonstrated that Lin28 binds to the 5'

stem or the loop region of pre-let-7 to block its processing by the Dicer ribonuclease

(Rybak et al., 2008) Heo et al (2008) observed that Lin28 promotes the terminal

uridylation of pre-let-7 in the cytoplasm, which undergoes degradation subsequently Furthermore, two studies identified Zcchc11 as the uridylyl transferase recruited by Lin28,

responsible for the uridylation of pre-microRNA in human and mouse (Heo et al., 2009; Hagan et al., 2009) Heo et al (2009) also pinpointed a specific tetra-nucleotide RNA

sequence motif (GGAG) in the terminal loop of pre-let-7 that is essential for recognition by LIN28’s CCHC-type zinc finger domains Thus, LIN28 may be able to interfere with both nuclear and cytoplasmic let-7 processing at multiple post-transcriptional levels during let-7 maturation

The repression of mature let-7 by LIN28 is important for pluripotency in human ES

cells let-7 family is well studied in C elegans, where let-7 regulates a set of target genes

through post-transcriptional repression to control the transition from undifferentiated,

proliferating stem cell to differentiated, quiescent cells (Büssing et al., 2008) Moreover,

let-7 was shown to inhibit High-mobility group AT-hook 2 (HMGA2) (Lee and Dutta

2008), RAS (Johnson et al., 2005), Myc (Shah, 2005) and cell-cycle genes (Johnson et al.,

2007), all of which play pivotal roles in ES cell renewal Although LIN28 is one of the four genes sufficient to accomplish reprogramming human fibroblast cells into iPS cells, it

is not indispensable Another known negative regulator of various let-7 family members,

MYC appears to be able to substitute for LIN28 (Lowry et al., 2008; Park et al., 2008),

which links their reprogramming capability to the repression of let-7

LIN28 and let-7 interaction is also suggested to be involved in primordial germ cell

(PGC) development (West et al., 2009) Modulation of mouse Lin28 expression level

during ES cell differentiation revealed its role in development of germ cells Lin28 influences PGC development through let-7-mediated effects on B lymphocyte induced maturation protein 1 (Blimp1), a key regulator of germ-cell commitment (Saitou, 2009) LIN28 was also proven to be associated with malignancies through repression of let-7

Dangi-Garimella et al (2009) showed that human LIN28 and let-7 are part of the

metastasis signaling Ectopic expression of mouse Lin28 suggested that it contributed to the malignant phenotype and moreover, let-7 loop mutant could abrogate such effect of Lin28 overexpression, indicating the involvement of let-7 in Lin28 regulation on

metastasis (Viswanathan et al., 2009)

1.4.2 LIN28 can regulate target genes post-transcriptionally

Besides repressing microRNA processing, Lin28 is also shown to bind target mRNA

to regulate translation In one mouse ES cell study, Cyclins A and B and cdk4 mRNAs are found in Lin28-containing ribonucleoprotein particles (RNPs), and their protein levels are changed in response to alteration of Lin28 expression Importantly, the author achieved the stimulation of translation of reporter genes by vectors containing 3' UTR of cyclin B mRNA (Xu et al, 2009) In addition to the key cell cycle regulatory genes, replication-dependent histone H2a was also identified as a target of Lin28, which underscored the importance of coordinated regulation of gene expression by Lin28 to promote proliferation (Xu and Huang, 2009) Mouse Lin28 associates with RNAs containing translation initiation complexes, in which the translation initiation factor eukaryotic translation

initiation factor 3 subunit b (eIF3b) interacts with Lin28 directly (Polesskaya et al., 2007)

Furthermore, Lin28 binds to Igf2 mRNA and increases the efficiency of its translation initiation involving eIF3b, and the process is essential for skeletal myogenesis Recently,

Qiu et al (2010) revealed human LIN28 stimulates OCT4 mRNA translation by recruiting

RNA helicase A (RHA), a component of translational machinery to facilitate RNP remodeling during translation in human ES cells RHA has been reported to be capable of promoting the formation of RNA-induced silencing complexes (RISC), which is linked with LIN28’s function in microRNA processing (Robb and Rana, 2007) Thus, LIN28 can selectively bind RNA substrates by recognition of binding sites and subsequently recruits eIF3b and RHA to regulate initiation and processing of translation of target mRNAs respectively

1.4.3 Overexpression and knockdown studies of LIN28

Modulation of gene expression by overexpression or knockdown has been commonly used to explore the target gene’s functions The first study of LIN28 in human ES cells was reported by Darr and Benvenisty (2008) In this study, clones stably overexpressing LIN28 were created, which formed around one third of the undifferentiated colonies that parental cells formed They found that the decrease in colonality was due to the slower rate in cell cycle (higher proportion in G1/G0 stage) and the increased differentiation to extra-embryonic endoderm However, knockdown of LIN28 by siRNA didn’t cause change of pluripotency status or cell cycle profile In human EC cell line PA-1, when LIN28 was down-regulated using siRNA, a decrease of 65% in cell viability could be observed (Peng

et al., 2009) The difference may be due to the higher knockdown efficiency in the latter

study In mouse ES cells, modulation of Lin28 didn’t cause differentiation, but overexpression and knockdown demonstrated that it promoted cell proliferation by

facilitating the progression from S to G2/M phase (Xu et al., 2009) This observation led

them to discover the post-transcriptional regulation on cell cycle related genes by Lin28

Heo et al (2009) found that knockdown of Lin28 caused lower Oct4 and Nanog mRNA

during EB formation, which implied the role of Lin28 in differentiation

Two studies employed stable transgenic mouse cell lines over-expression Lin28 to explore its function during differentiation Mouse ES cell line overexpressing Lin28 under the induction of doxycycline (DOX) was used to form EBs The overexpression was

accompanied by increase of primordial germ cells (PGCs) (West et al., 2009) In the

second study, constitutive expression of Lin28 in mouse P19 EC cells blocked glial differentiation but promoted neurogenesis, when cells were grown as aggregates with

retinoic acid (Balzer et al., 2010) Furthermore, they also developed various stable lines

overexpressing Lin28 but with mutants in its functional domains Through the comparison with the mutated versions, they discovered that the conserved domains were differentially required for the effect of Lin28 on cell fates

1.4.4 Objective of the current study

The transcriptome of human ES cells is very distinct from the rest of the cell and tissue types as revealed by a comparison of the human ES cells with different tissue types

We hypothesize that genes responsible for the maintenance of the pluripotent state share a common expression pattern and are significantly up-regulated in undifferentiated human

ES cells In addition, well-known pluripotency genes, such as POU5F1, SOX2 and NANOG, typically exhibit a gradual decrease in their gene expression profiles upon differentiation and as such are not really ideal markers for assessment of the differentiation status of human ES cells (Bhattacharya et al., 2005) Analysis of expression profile of undifferentiated, partially differentiated and differentiated human ES cells will be carried

out in this study to identify genes whose expression levels increase or decrease upon differentiation qRT-PCR will be used to verify their expression profiles using EBs harvested at different time points Though these, we aim to identify suitable gene markers that could indicate the differentiation status of human ES cells more appropriately In particularly, we would like to uncover gene markers that show a sharp decline in expression even at early stages of human ES cell differentiation

Reprogramming factor LIN28 appears to be of great importance in maintenance and

establishment of pluripotency in human ES cells (Richards et al., 2004; Yu et al., 2007; Heo et al., 2009; Hagan et al., 2009) However, various studies show that its knockdown

does not cause differentiation of ES cells directly (Viswanathan et al., 2008; Darr and Benvenisty, 2008; Xu and Huang, 2009; Balzer et al., 2010) This study seeks to elucidate LIN28’s functional role in human ES cell pluripotency siRNA- and shRNA-mediated LIN28 knockdown will be conducted in human embryonal carcinoma cell line, NCCIT, which will be used as an alternative model to human ES cells because of its resemblance to

human ES cells and its convenience for culture (Sperger et al., 2003; Boyer et al., 2005)

After the knockdown of LIN28, transcriptome analysis using Illumina DNA microarrays will be carried out to identify transcripts that are perturbed by the knockdown, and this will allow us to determine genes that are potentially regulated by LIN28 in the NCCIT cells

Chapter 2 Materials and Methods

2.1 Culture of human ES cell line

2.1.1 Preparation of feeder cells

J28 mouse embryonic fibroblast (MEF) cells were maintained in MEF medium which consisted of Dulbecco’s modified Eagle’s medium: high glucose (DMEM) supplemented with 10% One Shot™ Fetal Bovine Serum and 1% L-Glutamine (all from GIBCO/Invitrogen) Cells were passaged with 0.05% Trypsin/EDTA every 2 to 3 days, once confluency was reached At the 5th passage, MEFs were bulk cultured in T175 flasks, harvested and gamma irradiated with a dosage of 3000 rad Gamma irradiated MEFs were spun down and re-suspended in freezing medium which consisted of 90% One Shot™ Fetal Bovine Serum (FBS; GIBCO/Invitrogen) and 10% DMSO (sigma) Irradiated MEFs were stored in liquid nitrogen

2.1.2 Maintenance of human ES cells

The human ES cell line, HES3, from ES Cell International, Singapore (http://www.escellinternational.com), was cultured on irradiated MEFs in HES media [DMEM-F-12 supplemented with 20% knockout serum replacement (KSR), 2 mM non-essential amino acids (NEAA), 2 mM L-glutamine, and 4 ng/ml basic fibroblast factor (bFGF) (all from Invitrogen, Carlsbad, California, USA)]

2.1.3 Preparation of embryoid bodies

HES3 cell colonies were cultured until confluency, which was around day 6 or 7 Differentiated colonies observed under a Leica M28 dissecting microscope (Leica

Microsystems GmbH, Wetzler, Germany) were removed by scrapping with a sterile needle Thereafter, the human ES cells were detached by digestion with 0.5 ml collagenase IV (BD Biosciences, San Jose, CA, USA) per 35mm dish for 5 min The cells were then scraped off with a sterile plastic cell scraper and pipetted up and down several times to obtain small clumps of approximately 100-150 cells Cell clumps were centrifuged at 600 g for 2 minutes and resuspended in differentiation medium [Glasgow Minimum Essential Medium (Invitrogen) supplemented with 2 mM L-glutamine, 2 mM NEAA, 0.1 mM β-mercaptoethanol and 10% KSR] and seeded on ultra-low-adherence 6-well plates (Corning,

NY, USA) Medium was changed every two days and EBs were harvested at 12 h and 1, 3,

5, 7, and 14 days

2.2 Culture of NCCIT cell line

The pluripotent EC cell line, NCCIT was obtained from the ATCC (ATCC Number: CRL-2073) and maintained in Roswell Park Memorial Institute medium (RPMI 1640; GIBCO/Invitrogen) supplemented with 10% FBS (Hyclone/Thermo Fisher Scientific), 2

mM L-Glutamine and 50 U/ml penicillin / 50 µg/ml streptomycin (GIBCO/Invitrogen) Cells were passaged using 0.05% Trypsin EDTA every 3 to 4 days, according to the time

at which confluency was reached

2.3 Preparation of shRNA vectors targeting LIN28

Two shRNA target sequences of LIN28 were selected from the RNAi consortium (TRC) library database (http://www.broadinstitute.org/rnai/trc) shRNA sequences were ordered as individual oligonucleotides (Table 1) and annealed by heating to 95oC for 5 min

in a heat block, followed by slow cooling to room temperature by leaving on the benchtop Annealed oligonucleotides encoding respective shRNA were cloned into pLVTHM

linearized by ClaI and MluI digestion, downstream of the TetO-H1 region (Fig 4) The inserts were validated by NdeI digestion and sequenced using H1 primer: 5’

AGGAAGATGGCTGTGAGG 3’ To construct the inducible shRNA vector, the two

pLVTHM-LIN28 shRNA constructs were cut with MscI-FspI and the inserts containing the LTR/SIN were cloned into pLVET-tTR-KRAB (Szulc et al, 2006) plasmid restricted with MscI-FspI The correct clones were validated by MluI digestion

Figure 4 Construction of lentiviral inducible shRNA vectors targeting LIN28 shRNA insert

oligonucleotides were designed to have MluI and ClaI sticky overhangs and a diagnostic NdeI restriction

enzyme site after the terminating TTTTT sequence The annealed oligonucleotides were first cloned into

pLVTHM vector linearized by MluI and ClaI digestion Then the pLVTHM-LIN28 shRNA constructs were cut with MscI-FspI and the inserts containing the LTR/SIN were cloned into pLVET-tTR-KRAB vector linearized with MscI-FspI

Table 1 Oligonucleotides used in shRNA vector cloning

2.4 Transfection and lentivirus transduction of mammalian cells

2.4.1 Transfection of supercoiled shRNA vectors

A mixture of pLVET-LIN28sh1 and pLVET-LIN28sh2 vectors were used to transfect NCCIT cells to obtain synergistic and higher knockdown efficiency pLVET-tTR-KRAB, which does not contain any shRNA insert, was used as negative control NCCIT cells were seeded on a 6-well tissue culture plate (Nunc/Thermo Fisher Scientific) at a density of 8 X 105 cells per well Transfection was carried out the next day The shRNA vectors were transfected using the FuGENE HD transfection reagent (Roche) at a ratio of 2

µg of plasmid : 6 µl of FuGENE HD DNA was diluted in 100 µl of Opti-MEM® I Reduced Serum Medium (Invitrogen) Then FuGENE HD was added into the DNA diluent, after which the mixture was incubated for 15 min at room temperature The transfection complex was added to the cells in a drop-wise manner The transfection was repeated the next day and the same procedure was followed as above Culture medium was changed just before transfection and the following day after transfection DOX (500 ng/ml) induction was initiated two days post-transfection

2.4.2 Transfection of siRNA

siRNA specific to LIN28 (NM_024674) was obtained from Dharmacon TARGET plus SMARTpool), which contains a set of four different double-strand siRNA oligos Cy3 labelled negative control RNA (Qiagen) was included as well Lipofectamine RNAiMAX was used as transfection reagent, because it was proven to be more efficient than Lipofectamine 2000 and Oligofectamine in delivering siRNA into human ES cells

(ON-(Zhao et al., 2008) NCCIT cells were seeded on the 24-well plate at a density of 2X105

cells per well The transfection was performed on the following day using Lipofectamine RNAiMAX (Invitrogen) at a ratio of 60 pmol of siRNA pool : 2 µl of transfection reagent

siRNA and Lipofectamine were diluted in 50 µl of Opti-MEM® I Reduced Serum Medium (Invitrogen), respectively The diluted siRNA was combined with the diluted Lipofectamine RNAiMAX and was incubated for 20 min at room temperature before adding the siRNA-Lipofectamine RNAiMAX complex to the cells in a drop-wise manner The culture medium was changed and the transfection was repeated the next day as above The cells were harvested 48 or 72 hours after the first transfection

2.4.3 Lentivirus transduction of NCCIT

The conditional shRNA expression lentiviral vector pLVET-tTR-KRAB was chosen This “Tet-On” version vector contained a gene cassette encoding the tetracycline repressor (tetR) fused to Kruppel-associated Box gene (KRAB) The pLVET-tTR-KRAB-mediated repression of Pol II (EGFP) and Pol III (shRNA) promoters that were juxtaposed to the tet operator (tetO) sequences could be reversibly controlled by doxycycline (DOX)

simultaneously (Szulc et al., 2006) Thus cells that had stably integrated insert from the

vector construct could be enriched by selection of the EGFP marker Because it was demonstrated that multiple shRNAs targeting different regions of the same gene could

have synergistic RNAi effect to improve knockdown efficiency (Song et al., 2008),

lentiviral vectors pLVET-LIN28sh1 and pLVET-LIN28sh2 were mixed at the ratio of 1:1 and sent to Burnham Institute for Medical Research, Viral Vector Core Facility (La Jolla, California, USA) to package into lentiviral particles One day before transfection, the NCCIT cells were seeded on the 96-well tissue culture plate (Nunc) at a density of 25,000 cells per well Before transduction, the old medium was replaced by 100 µl fresh medium followed by 15 minutes incubation at 37 oC An aliquot of 100 µl lentiviral particles was added into the culture medium to transduce NCCIT cells at a multiplicity of infection (MOI) of 4 Polybrene (5 µg/ml) was used to increase the transduction efficiency The

NCCIT cells were incubated with lentiviral particles for 12 hours and then the medium was changed DOX (1 µg/ml) induction was initiated 36 hours post-transduction

2.4.4 Lentivirus transduction of HES3

To coat the 96-well tissue culture plate (Nunc), BD MatrigelTM hESC-qualified Matrix (BD Biosciences) was diluted in pre-chilled DMEM-F-12 (Invitrogen) at a ratio of 1:100 Incubation on ice for 10 min was carried out and 50 µl of the mixture was added to each well The plate was then incubated at room temperature for 1 hour The undifferentiated HES3 cells were passaged by mechanical cutting and seeded at a density

of 20,000 cells per well in mTeSR®1 medium (STEMCELL technologies, Vancouver, BC, Canada) On the following day, before transduction, old medium was replaced by 100 µl fresh mTeSR®1 medium followed by 15 minutes incubation at 37 0C An aliquot of 100 µl lentiviral particles was added into the culture well to transduce HES3 cells at a MOI of 5 Polybrene (5 µg/ml) was used to increase the transduction efficiency The HES3 cells were incubated with lentiviral particles for 12 hours and then the medium was changed DOX (1 µg/ml) induction was initiated 36 hours post-transduction

2.5 Fluorescence Activated Cell Sorting (FACS)

For NCCIT cells transfected by shRNA vectors or transduced by lentiviral particles, enhanced green fluorescent protein (EGFP)+ or EGFP- cells were enriched by FACS The NCCIT cells were washed once with PBS and incubated with 0.05% Trypsin/EDTA for 3-

5 min at 37 oC Culture medium was then added to stop digestion The cell suspension was centrifuged at 600 g for 2 minutes to pellet cells Next, the cells were resuspended in NCCIT culture medium at the concentration of around 1 million cells per ml and blasted into single cells Cell sorting was performed using the MoFlo sorter (Beckman Coulter)

Forward and side-scatter plots were used to exclude dead cells and debris from the histogram analysis Under the 488nm argon laser, the cells with fluorescence signal greater than 102 were collected as EGFP positive cells The NCCIT normal cells were used as negative control The analysis was performed using software Summit V4.5 (Dako Colorado, Inc Fort Collins, CO, USA)

2.6 SAGE data analysis

2.6.1 SAGE Libraries

The construction of SAGE libraries for undifferentiated HES3 and HES4 was

described earlier (Richards et al., 2004) Please note that the sequencing of SAGE tags was

extended to yield 192739 SAGE tags for HES3 Partially differentiated and differentiated HES3 SAGE libraries were generated from cells under prolonged high density cultures For these lab generated libraries, SAGE tag extraction was done using the SAGE2000 V4.5 software (Invitrogen), where the minimal ditag length were 24bp and 34

bp for shortSAGE and longSAGE library, respectively Database was managed with MS access and numerical analyses were preformed with MS Excel Cancer and normal tissue and EC cell libraries were downloaded from CGAP (http://cgap.nci.nih.gov/) In addition

to the four in-house human ES cell libraries mentioned above, nine longSAGE libraries

(H1P31, H1P54, HES3UD2, HES4UD2, H7, H9UD2, H13, H14 and BGO1) (Hirst et al.,

2007) and one shortSAGE H9UD1 (http://www.transcriptomes.org/) library were also used for the analysis The libraries used were listed in Table 2

Table 2 SAGE libraries used in this study

Tissue of origin  Library name  Tag length(bp)  Library